Landline/Cell Phone Dual Frame Reporting From the National Immunization Survey (NIS) and NIS-Teen, United States, 2010

Background: The National Immunization Survey (NIS) and NIS-Teen are random-digit-dialing (RDD) landline telephone surveys used to monitor vaccination coverage among children 19-35 months and adolescents 13-17 years, respectively, in the United States. There is potential for bias in reported vaccination coverage as the number of households using cell phones for most or all calls increases. Studies have shown that vaccination coverage estimates that include non-landline households might be lower than NIS estimates. In the fourth quarter of 2010, the NIS and NIS-Teen included an experimental cell-phone RDD sample to assess the potential bias.   

Objectives: Assess the impact on national vaccination coverage estimates using a landline frame and a land-line/cell phone dual frame.

Methods: The NIS collects data for children and teens utilizing a two-phase survey design with a RDD telephone survey of households followed by a Provider Record Check (PRC), a mail survey of the vaccination providers of those in the RDD survey. Weights for the landline and cell-phone samples were determined with adjustments for non-completion of the telephone-status screener for the cell-phone sample and for combining the samples

Results: CASRO response rates for cell-phone respondents were lower (NIS:25.9%, NIS-Teen:22.3%) than for landline respondents (NIS:63.8%, NIS-Teen:57.9%). The cell-phone sample included higher proportions of mothers reporting younger age, Hispanic or non-Hispanic black race/ethnicity, high school or less than high school education, and lower poverty status. NIS differences between landline single frame and landline/cell phone dual frame coverage estimates ranged from -1.2 to 2.2 percentage points and NIS-Teen differences ranged from -0.6 to 0.6 percentage points.

Conclusions: A land-line/cell phone dual frame can improve representativeness of the sample, but the effect of non-response in the cell phone population must be evaluated. Switching to a dual frame balances increased survey representativeness, decreased response rates, and increased cost.